Revert PR 4764 & 4813 related to R1 RoPE (#4959)

Author: guoyuhong

.github/workflows/pr-test-amd.yml

@@ -89,7 +89,6 @@ jobs:
           docker exec ci_sglang pip uninstall sgl-kernel -y || true
           docker exec -w /sglang-checkout/sgl-kernel ci_sglang bash -c "rm -f pyproject.toml && mv pyproject_rocm.toml pyproject.toml && python3 setup_rocm.py install"
           docker exec ci_sglang pip install -e "python[dev_hip]"
-          docker exec ci_sglang pip install py-spy || true
 
           docker exec -w / ci_sglang git clone https://github.com/merrymercy/human-eval.git
           docker exec -w /human-eval ci_sglang pip install -e .

python/sglang/srt/layers/rotary_embedding.py

@@ -651,18 +651,6 @@ def forward(
         query: torch.Tensor,
         key: torch.Tensor,
         offsets: Optional[torch.Tensor] = None,
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
-        if _is_cuda_available:
-            return self.forward_cuda(positions, query, key, offsets)
-        else:
-            return self.forward_native(positions, query, key, offsets)
-
-    def forward_native(
-        self,
-        positions: torch.Tensor,
-        query: torch.Tensor,
-        key: torch.Tensor,
-        offsets: Optional[torch.Tensor] = None,
     ) -> Tuple[torch.Tensor, torch.Tensor]:
         """PyTorch-native implementation equivalent to forward()."""
         query_rot = query[..., : self.rotary_dim]